A neural network was taught to reconstruct 3D images of complex materials
The use of microcomputed tomography is associated with some difficulties, in order to eliminate these problems scientists used the method of defect reconstruction, which is used in digital processing in the restoration of works of art. They proposed using generative adversarial networks (GANs) with 3D encoders and decoders to solve the problems of 3D microcomputed tomography image reconstruction. Material microstructure reconstruction is a start in the development of a fully automatic generative algorithm that will make it possible to create innovative materials with specific properties that will meet the requirements of applications. 
2021-08-19